Distance from barline to accidental is too small

This fixed the second measure in my example - there is no longer too *much* space - but the first case still shows the overlapping accidental, and an accidental *on* the staff is still too close to the barline. It's supposed to honor the "Barline to accidental distance" but it is not doing so. I appreciate that the whole point of the new layout is to allow more compact spacing. And I can see that at some level this is working as the algoriuthm is designed to allow - we are placing the accidental in space to the left of the note as long as it does not collide with anything. I guess maybe the way we set these distances using separate options for barline to note and barline to accidental distances could stand to be revisited.

But in any case, the results currently do not look good. The second note should not be so close, and I am not so sure about the first either:

One way or another, I think we still need to honor options to set the minimum distance from barline to note versus accidental, as suggested in Gould p. 42-43.

Current behavior is different but still not good:

No matter how I set the barline to accidental distance in style settings, I get the above, which is too much space. It looks like we are using the barline to note distance to set the distance to the accidental. Previously, as shown in #3 above, we were using barline to note distance to set the distance to the note and then squeezing the accidental in between the note and barline. That basic approach was more correct; we just should made sure to obey the barline to accidental distance as a minimum.

Current behavior for me is back to having the accidental too close to the barline by default as shown in https://musescore.org/en/node/106726#comment-482671. But on the positive side, the "Barline to accidental" setting in Style / General / Measure now appears to work as expected, which according to my comment it didn't before.

I would recommend increasing the default value from 0.3sp to somewhere between 0.5sp and 1sp. Gould (on page 43) gives 1sp as the best value normally, but suggests one can close it to as little as 0.5sp "where space is limited". We don't have an automatic way of adjusting based on this (that would be interesting!) so I'd recommend a compromise of maybe 0.75sp. I note that 2.x has this value set to 0.3sp, but somehow it resulted in more space. I think 2.x was adding some extra padding elsehwere, because a 0sp setting did not result in the accidental touching the barline as it does in master. So we might also need to do some fixup on import of older scores.

I do also think that even in the case where the first note has an accidental, we should still honor the "Note left margin" (which I guess is the new name for what was formerly "Barline to note distance", not sure that change was an improvement) as well as the "Barline to accidental distance". We did this in 2.x but are not currently doing it on master. Right now, if the note has an accidental, we honor "Barline to accidental distance" without question, which could potentially result in the note itself being closer than allowed "Note left margin" if the user increases the latter to a really large value. For example, here I have note left margin set to 5sp, barline to accidental distance set to 0.1sp for demonstration purposes:

There is no good reason the Ab in the first measure should not also be 5sp from the barline as the A is in the second. I suppose that's kind of nitpicking; if the user really wants a huge note left margin, he can also set an appropriate barline to accidental distance, and maybe having independent control isn't bad. Still, it is a change from 2.x

This was fixed years ago. Do you have a score where it still an issue? Note that invisible key signatures will mess that up, that's a separate issue.